Nowadays, both data statistics and analysis and AI model development need to take data as raw materials. Big data plays an important role in promoting economic development, social governance, and public management.
Against this backdrop, the underlying infrastructure of data circulation developed by Sudo based on privacy computing meet all needs in privacy computing scenarios. In combination with privacy computing technologies such as secure multi-party computation, federated learning, homomorphic encryption, and trusted execution environment, it provides distributed data fusion, joint modeling, and data usage of multi-party data for different organizations while also protecting data security and user privacy. The scheme completes the implementation of joint modeling and realizes multi-party data sharing without collecting or exchanging the original data of each other, making it possible for using and sharing big data and tapping to the value of data.
In the following project cases, through privacy computing technologies, Sudo empowers insurance companies and their partners to realize data collaboration based on bi-directional privacy protection and supports them in building big data foundational architecture.
Sudo uses privacy computing technologies to support Insurance companies and communication operators in their underlying insurance customers' operation based on secure data collaboration.
Before using privacy computing technology for data collaboration, data privacy was poorly protected in many aspects in the original collaboration. Specifically, both insurance companies and communication operators needed privacy protection. Insurance companies didn't want their own customer ID data and converted users to be retained and reversely inquired, and communication operators hoped that the data could be withheld in the database after business requirements were met.
Sudo provides a data collaboration solution based on privacy computing for communication operators Unit and insurance companies, effectively addressing their respective concerns about data interaction.
Insurance companies have established the positioning and mining model of the insurance intention groups through the ID number — phone number correlation graph and user portrait label.
First, Sudo helps both sides protect users' IDs (ID number in this case) through private set intersection (PSI) technology. Only insurance companies are able to know the matched users, while communication operators cannot retain or reversely attain users' identities. Only when insurance companies make it clear that consumer reach is required can the two parties exchange information. By doing so, insurance companies' concern about the ID leak was resolved.
Second, Sudo combines the labels of recorded product buyers of insurance companies with the portrait labels of users of communication operators through federated learning technology. This helps to establish an insurance intention model based on vertical federal learning. Such execution in cross-mining of data value will be achieved without exchanging original data by both parties.
By contrast, the insurance purchase interest mining model built based on Sudo's privacy computing technology evaluates the existing population data. The rate of insurance covered by the population ranking top 50 percent in the evaluation is 1.8 times the original performance.
By utilizing the privacy computing technology system developed proprietarily and privacy computing technologies including federated learning, Sudo helped clients in the insurance industry connect business that was supported by underlying data with that of their partner companies and realize data security sharing between subsidiaries and parent companies. On the premise that users' privacy is protected and data client users' information is available but invisible, we increased sample size and dimensions of data to help clients reach underlying customers efficiently and precisely predict labels of customer portrait, thus significantly improving the user conversion rate.
In addition, Sudo has helped a number of government agencies, central enterprises, state-owned enterprises, and large Internet companies to establish data collaboration. In advertising marketing, risk control, anti-fraud, anti-money laundering, data exchange, smart city, and other scenarios, Sudo has helped governments and enterprises collaborate on data of security and privacy.
Trailer of World Internet Conference
Copyright © World Internet Conference. All rights Reserved
Presented by China Daily. 京ICP备13028878号-23
Copyright © World Internet Conference. All rights Reserved Presented by China Daily. 京ICP备13028878号-23